Relative validity and utility of a short food frequency questionnaire assessing the intake of legumes in Scottish women

Transcription

1 RESEARCH PAPER Relative validity and utility of a short food frequency questionnaire assessing the intake of legumes in Scottish women A. Papadaki*, & J. A. Scottà *Preventive Medicine and Nutrition Clinic, Department of Social Medicine, University of Crete, Greece; Faculty of Medicine, Division of Developmental Medicine, Human Nutrition Section, University of Glasgow, UK; àdepartment of Nutrition and Dietetics, Flinders University, Adelaide, SA, Australia Correspondence Angeliki Papadaki, Preventive Medicine and Nutrition Clinic, Department of Social Medicine, Faculty of Medicine, University of Crete, Greece PO Box 2208, Heraklion , Crete, Greece Tel.: Fax: agpapadak@edu.med.uoc.gr Keywords food frequency questionnaire, legumes, relative validity, Scotland, soy foods. Abstract Objective To validate a five-item, semi-quantitative, short food frequency questionnaire (SFFQ) designed to estimate daily legume consumption over a week, against results obtained from 7-day food diaries (7-day FD). Design Participants completed a 7-day FD and at the end of this period completed the SFFQ, to indicate the number of times they ate five legume-containing dishes in the previous week and what size portion of each dish they consumed. Daily legume intake (g day )1 ) was calculated for both methods and participants were classified into tertiles of intake for each method. Subjects/setting Fifty-one healthy females aged years, employed at the University of Glasgow, Scotland, UK between May 2003 and December Results The two methods produced a similar mean intake of legumes [SFFQ: 14.8 (95% CI: ) versus 7-day FD: 14.9 (95% CI: ) g day )1 ] and the Pearson s correlation coefficient was (P = 0.038). Exact agreement within tertiles and gross misclassification were 54.9% and 9.8% respectively. The weighted kappa statistic indicated fair agreement between the two methods (j = 0.262). Conclusions The SFFQ is an acceptable instrument for estimating legume consumption over a week and can be used to rank individuals according to the intake of this food group in similar nutrition intervention studies. 467

2 468 A. Papadaki and J. A. Scott Conflict of interest, source of funding and authorship The authors declare that they have no conflict of interests. A. Papadaki was supported by a scholarship from the Greek State Scholarships Foundation (I.K.Y.). AP assisted with study design and conducted the study, analysed the data and co-authored the paper. JAS designed the study, assisted with data analysis and co-authored the paper. Introduction Legumes, or pulses, including peas, beans, lentils and chickpeas, are important sources of protein, dietary fibre, folate and various health-promoting phytochemicals (i.e. phytate, polyphenolics, flavonoids, indoles, lignans, saponins and trypsin inhibitors), while at the same time being a good source of a-linolenic acid and low in saturated fat (Anderson et al., 1999; Messina, 1999). Soy beans and soy products are also considered part of this food group and are high in monounsaturated fat and phytosterols (Anderson et al., 1999), in addition to being the major food source of the cancerprotective isoflavones genistein and daidzein (Steinmetz & Potter, 1996). The cholesterol-lowering, antioxidant and anticarcinogenic properties of legumes and soy foods have been demonstrated in ecological (Correa, 1981; Pisani et al., 1999) and prospective cohort studies (Ascherio et al., 1996; Fraser, 1999; Bazzano et al., 2001). The intake of soy foods has been associated with lower chronic disease incidence in Asian populations (Pisani et al., 1999; Nagata, 2000). Legumes have always formed an important component of the traditional Cretan Mediterranean diet and were consumed four times per week (Kafatos et al., 2000). Such frequency of legume consumption has been associated with a significant 22% reduction in risk of coronary heart disease and 11% reduction in risk of cardiovascular disease (Bazzano et al., 2001). In addition, a recent study has suggested that legumes are the most important dietary predictor of survival among older people, with a 7 8% reduction in mortality risk for every 20 g increase in daily intake (Darmadi-Blackberry et al., 2004). Most interventions promoting healthy eating to free-living individuals focus on increasing intake of fruits and vegetables and decreasing total and saturated fat intake. However, the number of interventions recommending regular consumption of legumes is limited (Leterme, 2002) and mainly confined to tertiary prevention trials (Renaud et al., 1995; Singh et al., 2002). In addition, it is usual practice in the United Kingdom (Williams, 1995) and the United States (Block et al., 1986) for legumes to be included in the fruit and vegetable category (Williams, 1995). In the traditional Mediterranean diet however, legumes were consumed frequently and independently of vegetables (Trichopoulou et al., 1995; Kafatos et al., 2000). Considering the potential health promoting benefits of legumes and soy foods (Messina, 1999; Bazzano et al., 2001; Darmadi-Blackberry et al., 2004) and their low consumption in Western countries (The Scottish Office, 1996; Gjonca & Bobak, 1997), the importance of the consumption of this food group, independently of vegetables, should be emphasized in future dietary intervention studies. For logistical reasons food frequency questionnaires (FFQs) are commonly used to evaluate the relationship between diet and chronic disease risk in large-scale, epidemiological studies. They can be self-administered, are less time consuming and less expensive compared with other dietary assessment methods, such as food records, dietary histories and 24-h recalls (Mc Keown et al., 2001; Van Assema et al., 2001). Two earlier studies on 20-item (Frankenfeld et al., 2003) and 40-item (Frankenfeld et al., 2002) FFQs assessing soy intake have reported good reliability and good-tomoderate validity when compared with plasma isoflavone concentrations and a more comprehensive FFQ that included questions about soy milk and tofu. Nevertheless, FFQs of such length are often impractical to use for dietary counselling and nutrition education programs, as they increase burden for participants (Kristal et al., 1990). The need for short, simple and valid instruments to rapidly estimate usual dietary intake for educational purposes has been well documented (Block et al., 2000; Wright & Scott,

3 Relative validity and utility of a SFFQ ). To date, studies which have reported on the relationship between legumes and chronic disease have assessed legume intake with multiple-item FFQs (Mills et al., 1989; Ascherio et al., 1996; Fraser, 1999; Hu et al., 1999; Bazzano et al., 2001) designed to measure intake of a number of food groups in addition to legumes. To the best of our knowledge this is the first study to pilot test and validate a short food frequency questionnaire (SFFQ) specifically designed to measure legume intake. The aim of the present study was to assess the relative validity of a five-item, semi-quantitative, SFFQ designed to estimate daily legume, in addition to soy product, consumption over a period of a week, against the results obtained from 7-day estimated food diaries (FD). Such a questionnaire could be a valuable tool for assessing intake and measuring dietary change in primary prevention studies promoting established healthy eating patterns, such as the Mediterranean and Asian diets, to Western populations. Materials and methods Participants Female employees of the University of Glasgow were recruited as part of a 9-month controlled study investigating the effect on dietary behaviour of an internet-based, tailored nutrition intervention, promoting key aspects of the traditional Mediterranean diet, including regular legume consumption (at least 1 2 times week )1 ) (Papadaki & Scott, 2005). Eligibility criteria included females aged years, with internet and access at work and who were born or had lived in Scotland for more than 15 years. Participants were ineligible if they had a self-reported history of cardiovascular disease, cancer, diabetes or hypertension. We did not include participants with dietrelated health problems as it was thought that pre-existing treatments and medications might obscure the impact evaluation of the intervention. Recruitment strategies included advertisements in newsletters, flyers, postings on the University intranet and advertisements. Those who met the study criteria were sent an information sheet that described the study procedure in detail. Once their willingness to participate was verified, participants signed an informed consent form. Participation was voluntary and anonymity was ensured. A description of the intervention methodology and results of the impact and process evaluation have been published in detail elsewhere (Papadaki & Scott, 2005, 2006). The study was conducted from May 2003 to December 2004 and was approved by the Faculty of Medicine Ethics Committee of the University of Glasgow. Study design and procedures Seventy-five women showed interest in the original study (Papadaki & Scott, 2005). Of these, ten were ineligible, whereas 57 accepted and were eligible to take part. Four women withdrew prior to baseline measurements, resulting in 53 women actively participating at baseline. Fifty-one of these women completed both a 7-day estimated food diary and SFFQ at this time point. At the end of the week when completed diaries were returned to the researcher, participants were asked to complete the self-administered legume SFFQ. Having participants to complete the SFFQ on-site following the handover of the completed food diary prevented participants from cross-checking their SFFQ answers with their already completed diary. Short food frequency questionnaire A short, semi-quantitative FFQ was developed to assess intake of legumes and soy products over the previous week (Appendix). The SFFQ was selfadministered and took less than 10 min to complete. It requested participants to indicate the number of times they ate five legume containing dishes/meals (baked beans, legume soups, legumes added to mixed dishes, as well as soy beans/tofu and soy/veggie burgers) in the previous week and note the number of a priori standard-size servings they consumed. The legume food items in the SFFQ were those reported to be commonly consumed in Scotland (Food Standards Agency, 2002). Weekly frequency of intake was measured on an 8-point scale (from never to 7 days week )1 ) and the number of servings consumed on a 5-point

4 470 A. Papadaki and J. A. Scott scale (from none, to >3 servings ). The SFFQ provided participants with the opportunity to report their exact number of servings, if that was more than three. Servings were described in household units, e.g. half cup of cooked legumes, one cup of soup, one burger etc. For each food item in the SFFQ, servings consumed daily were calculated and multiplied by a nominal serving weight to generate a g day )1 figure. Serving weights were considered a priori to be equal to 100 g (one cup/small can) of baked beans, 200 g (one bowl) of soup, 120 g (medium piece) of burger/tofu and 60 g (½ cup) of legumes in the remaining food items. One serving of soup represented a nominal serving of 60 g of legumes. Total daily intake of legumes was calculated by adding the daily intake (in grams) for each food item in the SFFQ. Peas were assessed as part of this questionnaire, as in the original study that described the traditional Greek diet peas were included in the legume, and not the vegetable, food group (Trichopoulou et al., 1995). Reference instrument A 7-day estimated food diary (household measure technique) was used as the reference instrument to validate the legume SFFQ. This method is widely used as an acceptably valid and reliable tool to assess usual dietary intake and is one of the preferred methods for validating FFQs, following the weighed inventory (Cade et al., 2002). Participants were asked to record, in as much detail as possible, all foods and beverages consumed over a 7-day period and describe their food servings in household units (e.g. slices, cups, spoons) or provide weights when known (e.g. packaged foods). Detailed written instructions for completion of the FD were provided and participants were encouraged to contact the researcher if they had any questions regarding this procedure. FD were posted to participants, a date for commencement of diary completion was set and completed diaries were returned during a meeting with the researcher, one day following the last diary entry date (Papadaki & Scott, 2005). Dietary data from the 7-day food diaries (7-day FD) were checked for completeness and participants were contacted individually in order to clarify any queries (e.g. unknown food items or food servings). To avoid coding errors, data were entered into the computer and analysed by the same researcher, using the Diet5 program (Diet5 for Windows 2000; Robert Gordon University, Aberdeen, UK). In the case when a consumed food item was not included in the Diet5 program, participants were asked to provide specific recipes or food brand names, and in the latter case, this was substituted with an item of similar nutrient content. Mean daily individual legume and soy food intakes (g day )1 ) were calculated from the diaries. A food diary was to be excluded from further analyses if a participant s energy intake was <600 kcal ( kj) day )1 and >3500 kcal ( kj) day )1. Based on this criterion, none of the FD were removed from the analyses. Statistical analysis All analyses were performed using the Statistical Package for the Social Sciences (SPSS for Windows, release 11.5, 2002; SPSS, Chicago, IL, USA). Legumes and soy foods were considered as one food group for all analyses, reported here as legumes, and significance was defined as P < 0.05 (two-sided). Descriptive statistics were used to report baseline and sociodemographic characteristics of participants. Mean (with standard deviations) legume intakes were calculated from the SFFQs and the 7-day estimated FD and the Wilcoxon Signed Ranks Test was used to detect differences between the two dietary assessment methods. Relative validity of the SFFQ was determined by comparing legume intakes derived from the SFFQ with the results obtained from the FD. Data on mean legume intake (g day )1 ) were recoded into tertiles, separately for each method, in order to examine participants cross-classification and determine the percentage of participants classified into the same, adjacent and opposite tertiles. Data were checked for distribution and comparison of intakes (g day )1 ) and agreement within categories between the two dietary assessment methods were assessed using the Pearson s R correlation coefficient and the weighted kappa (j) statistic respectively.

5 Relative validity and utility of a SFFQ 471 Table 1 Sociodemographic characteristics of participants In dietary validation studies, desirable correlation coefficients tend to be higher than 0.5 (Willett, 1998) and it has been suggested that j values from 0 to 0.20 indicate poor agreement, 0.21 to 0.40 fair agreement, 0.41 to 0.60 moderate agreement, 0.61 to 0.8 good agreement and 0.81 to 1 very good agreement (Altman, 1992). Bland and Altman plots were also used (Altman & Bland, 1994). Results n =51 % Age (years) < > Ethnicity Caucasian Level of education Technical training University or higher Occupation Managerial and professional Intermediate Routine and manual Fifty-one Caucasian participants completed both dietary assessment methods at baseline. Participants age ranged from 25 to 55 years, with a mean age of 40.3 years (SD 7.3). A summary of sociodemographic characteristics of participants is presented in Table 1. Sixteen and twenty-one participants reported eating no legumes when completing a food diary and SFFQ respectively. In addition, 32 participants who reported eating legumes with one dietary assessment method also reported consumption when assessed with the other method. Legume intake mostly resulted from consumption of baked beans (52% of consumption), soups (33% of consumption) and legumes used in mixed dishes (e.g. stews and dips) (15% of consumption). None of the participants reported eating soy beans or tofu with any of the methods. In addition, none of the participants were vegetarians. The two dietary assessment methods produced a similar mean daily intake of legumes [SFFQ: 14.8 (95% CI: ) versus 7-day FD: 14.9 (95% CI: ) g day )1 ; P = 0.862]. Correlations and measures of agreement between the results of the SFFQ and the 7-day estimated FD are listed in Table 2. The correlation between the two measures of total daily legume intake generated by each method was lower than the desirable 0.5 (Willett, 1998) (r = 0.353, P = 0.038). The proportion of participants classified by the SFFQ and the FD into the same tertile for legume intake was 54.9%, whereas gross misclassification (into opposite tertiles) was 9.8%. The weighted j was 0.262, indicating fair agreement between the two dietary assessment methods. A Bland Altman plot was carried out for crude legume intake (Fig. 1). This plot displays the difference between two methods against their mean and assesses the relationship between the measurement error and the true value (Bland & Altman, 1986). Calculation of differences between the two methods showed that the mean value was close to zero. Slightly fewer (94.1%) than 95% of the values were within 2 SD (349.3, )350.7) above or below this value. Table 2 Correlation and agreement within tertiles between the short food frequency questionnaire (SFFQ) and 7-day estimated food diaries Pearson s R correlation coefficient Weighted kappa Similarly classified (%) Adjacently classified (%) à Dissimilarly classified (%) n = * *P = Percentage of participants classified into the same tertile of the distribution by the 7-day food diary and the SFFQ. à Percentage of participants classified into adjacent tertiles of the distribution by the 7-day food diary and the SFFQ. Percentage of participants classified into the opposite tertile of the distribution by the 7-day food diary and the SFFQ.

6 472 A. Papadaki and J. A. Scott Difference in legumes (g) (7dFD-SFFQ) Mean legume intake by two methods Figure 1 Bland and Altman plot for legume intake as estimated by a short food frequency questionnaire and a 7-day estimated food diary. Discussion In this study, the relative validity of a short, semiquantitative FFQ estimating legume and soy food consumption over a week was examined among Scottish women aged years. The present study showed that this legume SFFQ provided a fair estimate of legume and soy food intake over this period, in relation to the standard reference instrument of 7-day estimated FD (Altman, 1992). In addition, gross misclassification was less than 10% and more than half of participants were classified into the same tertile of the distribution by the two dietary assessment methods, suggesting that the SFFQ can provide reasonable rankings of individuals on the basis of legume and soy food intake. The SFFQ used in this study estimated legume intake over a 1-week period. Although many FFQs are used to estimate dietary intake over longer periods (e.g. previous year), particularly when designed to estimate several food groups and/or nutrients or estimate consumption of items with seasonal variation, the longer reference time frame of such instruments might result in memory-related problems of recall, which in turn will affect the validation of FFQs. In the present study, it was not expected that legumes would be affected by seasonal variation in dietary intake. In addition, both dietary assessment methods in this study estimated legume consumption over the same time period (previous 7 days), which increases the validity of results (Cade et al., 2002). However, future validation of the SFFQ with a longer reference period, such as 1 year, against multiple FD or multiple 24-h recalls might be useful to confirm the representativeness of the results and the potential ability of this SFFQ to assess habitual legume consumption, in addition to shorter-term intake. Two earlier studies have validated soy (but not legume) intake FFQs in Western populations. The FFQs used in these studies showed moderate validity among post-menopausal women (Frankenfeld et al., 2003) and good agreement among male and female adults (j = 0.75) (Frankenfeld et al., 2002), compared with a comprehensive FFQ that included questions on soy food intake. Both these studies measured plasma concentrations of daidzein and genistein as biomarkers of soy intake and found significant correlations, slightly higher with the soy FFQ, compared with the comprehensive FFQ, between estimates of dietary intake and plasma values [r = 0.45 and 0.53 (Frankenfeld et al., 2002) and r = 0.37 and 0.43 (Frankenfeld et al., 2003) for daidzein and genistein respectively]. Direct comparisons with the present study are difficult to undertake however, due to different

7 Relative validity and utility of a SFFQ 473 methodologies and because our study is the first, to our knowledge, to validate a SFFQ estimating legume, in addition to soy food, consumption against the results obtained from 7-day estimated FD. Although the results from this study are promising, our sample consisted of a self-selected group of relatively well-educated women, who might have responded differently to the SFFQ and might have provided more accurate responses than nonvolunteers (Cade et al., 2002). It is noteworthy that participants in this study had, at baseline, a more favourable average daily consumption of legumes compared with Scottish women s (consumers, aged years) median intakes (15 g day )1 versus 2 g day )1 ) (Papadaki & Scott, 2005). However, their intakes were still lower compared with the median intakes ( 49 g day )1 ) of elderly Greek women consuming a traditional Greek diet (Trichopoulou et al., 1995). Although the sample size was acceptable for a validation study (Cade et al., 2002), further evaluation among a more diverse sample which includes males and takes into consideration ethnic, socioeconomic, educational and other demographic characteristics will be useful. A further limitation of this study is that we did not determine the reliability of the SFFQ, by measuring its external consistency (repeatability). This was not considered appropriate because the intervention immediately following the baseline administration of the SFFQ (Papadaki & Scott, 2005) might have triggered participants to change their legume intake, therefore affecting the results. Future evaluation of the SFFQ for possible use in epidemiological, in addition to intervention studies, should therefore optimally include a test retest reliability measurement, taking into account the potential variation in intake. It addition, it has been well documented that in validation studies, the test instrument should be administered prior and independently, of the reference measure, to avoid any training effect of diet recording and providing answers based on the reference instrument (Cade et al., 2002). Although the SFFQ in the present study was completed independently of the 7-day estimated food diary, its administration followed the food diary completion. This was because validating the SFFQ was a secondary aim of the original study (Papadaki & Scott, 2005), where the 7-day estimated food diary was the main dietary assessment method to evaluate effects of the intervention. It was intended that if the SFFQ provided a good estimate of legume intake, once validated against the 7-day FD, it would be designed as a web-based questionnaire to be completed by participants online during the course of the study, in order to provide tailored educational feedback and assist in setting and achieving dietary goals regarding legume intake. In future validation studies the SFFQ should be ideally administered prior to the assessment of the reference instrument. In addition, improved definition of serving size and perhaps the use of food models/pictures to aid participants with reporting more accurate consumptions would possibly improve the agreement between the two assessment methods. This paper reports the results of the validation of a self-administered, SFFQ developed to assess legume intake in Western populations. Brief instruments measuring legume consumption that are quick to complete and simple and inexpensive to administer can be important tools for informing targeted feedback interventions and for the evaluation of nutrition education programmes targeted at increasing intake of this food group. Because this SFFQ is brief, can be self-administered and is easily scored, it offers an inexpensive way to provide dietary assessment and educational feedback. Although the limitations of this study make the generalizability of the results difficult, the SFFQ provides a fair estimate of legume consumption and is acceptable for assessing intake of this food group and classifying individuals into categories of intake in similar populations and similar studies as the one here. The relative validity and reliability of the instrument require further evaluation amongst males and other cultural groups before it can be adopted for widespread use in future intervention studies. References Altman, D. (1992) Practical Statistics for Medical Research. London: Chapman & Hall.

8 474 A. Papadaki and J. A. Scott Altman, D. & Bland, J. (1994) Quartiles, quintiles, centiles, and other quantiles. Br. Med. J. 309, 996. Anderson, J.W., Smith, B.M. & Washnock, C.S. (1999) Cardiovascular and renal benefits of dry bean and soybean intake. Am. J. Clin. Nutr. 70, 464S 474S. Ascherio, A., Hennekens, C., Willett, W.C., Sacks, F., Rosner, B., Manson, J.A., Witteman, J. & Stampfer, M.J. (1996) Prospective study of nutritional factors, blood pressure and hypertension among US women. Hypertension 27, Bazzano, L.A., He, J., Ogden, L.G., Loria, C., Vupputuri, S., Myers, L. & Whelton, P.K. (2001) Legume consumption and risk of coronary heart disease in US men and women. Arch. Intern. Med. 161, Bland, J. & Altman, D. (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 8, Block, G., Hartman, A.M., Dresser, C.M., Carroll, M.D., Gannon, J. & Gardner, L. (1986) A data-based approach to diet questionnaire design and testing. Am. J. Epidemiol. 124, Block, G., Gillespie, C., Rosenbaum, E.H. & Jenson, C. (2000) A rapid food screener to assess fat and fruit and vegetable intake. Am. J. Prev. Med. 18, Cade, J., Thompson, R., Burley, V. & Warm, D. (2002) Development, validation and utilisation of food-frequency questionnaires-a review. Public Health Nutr. 5, Correa, P. (1981) Epidemiological correlations between diet and cancer frequency. Cancer Res. 41, Darmadi-Blackberry, I., Wahlqvist, M.L., Kouris-Blazos, A., Steen, B., Lukito, W., Horie, Y. & Horie, K. (2004) Legumes: the most important dietary predictor of survival in older people of different ethnicities. Asia Pac. J. Clin. Nutr. 13, Food Standards Agency (2002) The Dietary and Nutritional Survey of British Adults. London: HMSO. Frankenfeld, C.L., Patterson, R.E., Kalhorn, T.F., Skor, H.E., Howald, W.N. & Lampe, J.W. (2002) Validation of a soy food frequency questionnaire with plasma concentrations of isoflavones in US adults. J. Am. Diet. Assoc. 102, Frankenfeld, C.L., Patterson, R.E., Horner, N.K., Neuhouser, M.L., Skor, H.E., Kalhorn, T.F., Howald, W.N. & Lampe, J.W. (2003) Validation of a soy food-frequency questionnaire and evaluation of correlates of plasma isoflavone concentrations in postmenopausal women. Am. J. Clin. Nutr. 77, Fraser, G.E. (1999) Associations between diet and cancer, ischemic heart disease and all-cause mortality in non-hispanic white California Seventh-day Adventists. Am. J. Clin. Nutr. 70, 532S 538S. Gjonca, A. & Bobak, M. (1997) Albanian paradox, another example of protective effect of Mediterranean lifestyle? Lancet 350, Hu, F.B., Rimm, E., Smith-Warner, S.A., Feskanich, D., Stampfer, M.J., Ascherio, A., Sampson, L. & Willett, W.C. (1999) Reproducibility and validity of dietary patterns assessed with a food-frequency questionnaire. Am. J. Clin. Nutr. 69, Kafatos, A., Verhagen, H., Moschandreas, J., Apostolaki, I. & Van Westerop, J.J.M. (2000) Mediterranean diet of Crete: foods and nutrient content. J. Am. Diet. Assoc. 100, Kristal, A.R., Abrams, B.F., Thornquist, M.D., Disogra, L., Croyle, R.T., Shattuck, A.L. & Henry, H.J. (1990) Development and validation of a food use checklist for evaluation of community nutrition interventions. Am. J. Public Health 80, Leterme, P. (2002) Recommendations by health organizations for pulse consumption. Br. J. Nutr. 88, S239 S242. Mc Keown, N.M., Day, N.E., Welch, A.A., Runswick, S.A., Luben, R.N., Mulligan, A.A., Mc Taggart, A. & Bingham, S.A. (2001) Use of biological markers to validate selfreported dietary intake in a random sample of the European Prospective Investigation into Cancer United Kingdom Norfolk cohort. Am. J. Clin. Nutr. 74, Messina, M.J. (1999) Legumes and soybeans: overview of their nutritional profiles and health effects. Am. J. Clin. Nutr. 70, 439S 450S. Mills, P.K., Beeson, W.L., Phillips, R.L. & Fraser, G.E. (1989) Cohort study of diet, lifestyle and prostate cancer in Adventist men. Cancer 64, Nagata, C. (2000) Ecological study of the association between soy product intake and mortality from cancer and heart disease in Japan. Int. J. Epidemiol. 29, Papadaki, A. & Scott, J.A. (2005) The Mediterranean Eating in Scotland Experience project: evaluation of an internet-based intervention promoting the Mediterranean diet. Br. J. Nutr. 94, Papadaki, A. & Scott, J.A. (2006) Process evaluation of an innovative healthy eating website promoting the Mediterranean diet. Health Educ. Res. 21, Pisani, P., Parkin, D.M., Bray, F. & Ferlay, J. (1999) Estimates of the worldwide mortality from 25 cancers in Int. J. Cancer 83, Renaud, S., de Lorgeril, M., Delaye, J., Guidollet, J., Jacquard, F., Mamelle, N., Martin, J.L., Monjaud, I., Salen, P. & Toubol, P. (1995) Cretan Mediterranean diet for prevention of coronary heart disease. Am. J. Clin. Nutr. 61, 1360S 1367S. Singh, R.B., Dubnov, G., Niaz, M.A., Ghosh, S., Singh, R., Rastogi, S.S., Manor, O., Pella, S.D. & Berry, E.M. (2002) Effect of an Indo-Mediterranean diet on progression of coronary artery disease in high risk patients (Indo- Mediterranean Diet Heart Study): a randomised singleblind trial. Lancet 360, Steinmetz, K.A. & Potter, J.D. (1996) Vegetables, fruit and cancer prevention: a review. J. Am. Diet. Assoc. 96,

9 Relative validity and utility of a SFFQ 475 The Scottish Office (1996) Eating for Health: A Diet Action Plan for Scotland, pp Edinburgh: HMSO, The Scottish Office Department of Health. Trichopoulou, A., Kouris-Blazos, A., Wahlqvist, M.L., Gnardellis, C., Lagiou, P., Polychronopoulos, E., Vassilakou, T., Lipworth, L. & Trichopoulos, D. (1995) Diet and overall survival in elderly people. Br. Med. J. 311, Van Assema, P., Brug, J., Ronda, G. & Steenhuis, I. (2001) The relative validity of a short Dutch questionnaire as a means to categorize adults and adolescents to total and saturated fat intake. J. Hum. Nutr. Diet. 14, Willett, W. (1998) Nutritional Epidemiology, 2nd edn. Oxford: Oxford University Press. Williams, C. (1995) Healthy eating: clarifying advice about fruit and vegetables. Br. Med. J. 310, Wright, J.L. & Scott, J.A. (2000) The fat and fibre barometer, a short food behaviour questionnaire: reliability, relative validity and utility. Aust. J. Nutr. Diet. 57, Appendix Legume short food frequency questionnaire How many days did you eat the following foods last week? (Please tick Ö one answer in every line. Include peas, canned or dried baked beans, lentils, chick peas, split peas etc.) Baked beans (on toast, side dish, etc.) Legume soups (e.g. lentil or split pea soup, etc.) Legumes added to mixed dishes (stews, pies, casseroles etc.) Soy beans or tofu Soy/veggie burgers Please check that you have ticked Ö every line. Number of days week ) How many servings of the following foods did you eat on a day in which you ate these foods last week? (Please tick Ö one answer in every line) A serving is: 1 cup baked beans ½ cup of cooked legumes or soy beans (dried or canned, drained) 1 bowl of legume soup 1 medium soy burger and 1 medium piece of tofu Baked beans (on toast, side dish, etc.) Legume soups (e.g. lentil or split pea soup, etc.) Legumes added to mixed dishes (stews, pies, casseroles etc.) Soy beans or tofu Soy/veggie burgers Please check that you have ticked Ö every line. Number of servings per day None >3, please specify